Congratulations are due once again to the Department of Homeland Security for engaging in open dialogue about its programs, even controversial ones like “E‑Verify” — a system that Congress may require all U.S. employers to use for running federal background checks on every single new employee.

Openness is healthy, and the comments to a recent post on E‑Verify by my old friend DHS Assistant Secretary for Policy Stewart Baker are poking some holes in his somewhat facile analysis. I’ll weigh in with a little more, based mostly on my recent paper “Electronic Employment Eligibility Verification: Franz Kafka’s Solution to Illegal Immigration.”

Baker says that critics claim the error rate in E‑Verify is as high as 4% and will lead to millions of Americans losing their jobs by mistake. To refute this, he points to a study commissioned by the Department of Homeland Security showing that 94.2% of new hires in a sample of 1,000 E‑Verify queries were automatically verified, 0.5% resolved a mismatch, and 5.3% received a final nonconfirmation (that is, they either didn’t try or couldn’t challenge the finding that they were ineligible for employment under U.S. immigration law).

Unfortunately, Baker doesn’t point to the actual study. He just links to a picture of a conclusion from it, so we can’t do much to analyze these figures. If these are the results from reviewing only 1,000 new hires by current E‑Verify users, that is far too small a sample and too skewed a group to reflect what would happen were the program taken national.

And he concludes: “Of the thousand, 942 are instantly verified. Instant verification of legal workers surely can’t be an error.” Of course it can! Any number of the 942 might have been illegal immigrants who submitted the name and Social Security Number of a legal worker to the employer.

But putting Baker’s glib, erroneous conclusion aside, I believe the 4% figure cited by critics is not about today’s small E‑Verify program. It’s the error rate in the Social Security Administration’s Numident database found by the SSA’s own Inspector General (and it’s 4.1%!). Simple math suggests that this would produce a tentative nonconfirmation in 1 out of 25 new hires in the country were E‑Verify to go national.

In fairness, that simple math may actually be simplistic — perhaps some cohorts have higher error rates and others lower. We know, for example, that naturalized citizens suffer error rates in the area of 10%. Perhaps older citizens that are leaving the workforce have higher error rates, leaving a lower error rate among current workers. And over time, the error rate would drop as workers were sent from their jobs to Social Security Administration offices trying to get their paperwork in order. (Put aside for now that the SSA takes more than 500 days to issue disability rulings.)

Baker’s conclusion that the 5.3% of workers finally nonconfirmed are illegal workers is without support. The statistic just as easily could show that the 5.3% of law-abiding American-citizen workers are given tentative nonconfirmations, and they find it impossible to get them resolved. More likely, some were dismissed by employers, never informed that there was a problem with E‑Verify; some didn’t have the paperwork, the time, or the skills to navigate the bureaucracy; and some were illegal workers who went in search of work elsewhere, including under the table.

American workers pushed out of the workforce by E‑Verify — Baker treats it as “common sense” that they’re illegal aliens, and he doesn’t look any further. The E‑Verify program does the same — it has no system for contesting or appealing final nonconfirmations.

With his post, Secretary Baker has only raised the question of error rates in E‑Verify. There are many sources of error in a system like this, and making it bigger would reveal more. Just because you have a glass coffee table, that doesn’t mean you can build a glass sundeck.

And we shouldn’t take our eye off the ball. “Mission creep” is a governmental law of gravity. Once in place, a national E‑Verify system would be used to give the federal government direct regulatory control over law-abiding Americans. Federal authorities would use it to control not just work, but housing, financial services, and access to alcohol, tobacco, and firearms — for starters. Secretary Baker himself recently suggested using a national ID to control our access to cold medicine. The list of things his successors might do is endless.